- Size
- 27.8B parameterscounted from the safetensors header on the Hub
- Licence
- custom, read it, Apache License 2.0Our reading, check the licence: A custom licence we have not read for this model: other makers use the same name for different texts. Read the model's own licence file before you use it.Licence text
- On the Hub since
- 21 Apr 2026the date the repo was created
- Context length
- 256k tokensfrom the model's config
- Kind
- Base model (no base model declared on the Hub)Qwen3_5ForConditionalGeneration
- On Hugging Face
- Qwen/Qwen3.6-27B2.5M downloads in the last 30 days, 2.4k likes
Scores
Epoch's name: qwen3.6-27b. Best closed model on this set: claude-fable-5-2_max, 95.1% (Anthropic). All 127 results
Epoch's name: qwen3.6-27b. Best closed model on this set: gpt-6.2-astra_max, 96.8% (OpenAI). All 341 results
- effort none7.0%±2.627 Aug 2026, log
Epoch's name: qwen3.6-27b. Best closed model on this set: gpt-6.2-astra_max, 95.0% (OpenAI). All 148 results
Epoch's name: qwen3.6-27b. Best closed model on this set: claude-fable-5_high, 100.0% (Anthropic). All 318 results
- effort none0.9%via Epoch AI
Epoch's name: qwen3.6-27b. Best closed model on this set: gemini-4-carbon_high, 36.0% (Google DeepMind). All 271 results
- default setting78.1%conceptualreasoning.ai
Epoch's name: qwen3.6-27b. Best closed model on this set: claude-fable-5-2_unknown, 99.2% (Anthropic). All 241 results
- default setting34.5%30.0 to 39.0conceptualreasoning.ai
Epoch's name: qwen3.6-27b. Best closed model on this set: claude-fable-5-2_xhigh, 70.6% (Anthropic). All 203 results
Claimed by the maker Claimed
The model card has no claimed scores in a machine-readable form (the card's model-index), so none are shown. The maker may give scores in the card text: read the card.
Other views, which we link to and do not copy: look this model up on Artificial Analysis and the Arena leaderboard.
Independent scores from Epoch AI (CC-BY 4.0), fetched 19 Nov 2026: Epoch AI, ‘Capabilities & benchmarking’. Published online at epoch.ai. Retrieved from ‘https://epoch.ai/benchmarks’ [online resource]. Each row links its own source. What the badges mean.
Run it yourself
Memory needed = weights + context cache + 0.50 GB for the runtime. Context cache at 8k tokens: 0.50 GB, worked out from the model's config (Qwen/Qwen3.6-27B), at 16-bit precision. Choose a machine above to see a verdict for each format. Method
GGUF for llama.cpp, Ollama, LM Studio and others
unsloth/Qwen3.6-27B-GGUF (converted by unsloth), 242k downloads. Sizes are the real files on the Hub, all parts added up. For image input you also need the projector file mmproj-F16.gguf (0.86 GB), not counted below.
| Format | Size | Memory needed | Get it |
|---|---|---|---|
| UD-IQ3_XXS | 11.2 GBfile size | 12.2 GB11.2 + 0.50 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.6-27B-GGUF:UD-IQ3_XXSOllama ollama run hf.co/unsloth/Qwen3.6-27B-GGUF:UD-IQ3_XXSDownload hf download unsloth/Qwen3.6-27B-GGUF --include "Qwen3.6-27B-UD-IQ3_XXS.gguf" --local-dir Qwen3.6-27B-GGUF |
| Q3_K_S | 11.5 GBfile size | 12.5 GB11.5 + 0.50 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.6-27B-GGUF:Q3_K_SOllama ollama run hf.co/unsloth/Qwen3.6-27B-GGUF:Q3_K_SDownload hf download unsloth/Qwen3.6-27B-GGUF --include "Qwen3.6-27B-Q3_K_S.gguf" --local-dir Qwen3.6-27B-GGUF |
| Q3_K_M | 12.7 GBfile size | 13.7 GB12.7 + 0.50 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.6-27B-GGUF:Q3_K_MOllama ollama run hf.co/unsloth/Qwen3.6-27B-GGUF:Q3_K_MDownload hf download unsloth/Qwen3.6-27B-GGUF --include "Qwen3.6-27B-Q3_K_M.gguf" --local-dir Qwen3.6-27B-GGUF |
| UD-Q3_K_XL | 13.5 GBfile size | 14.5 GB13.5 + 0.50 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.6-27B-GGUF:UD-Q3_K_XLOllama ollama run hf.co/unsloth/Qwen3.6-27B-GGUF:UD-Q3_K_XLDownload hf download unsloth/Qwen3.6-27B-GGUF --include "Qwen3.6-27B-UD-Q3_K_XL.gguf" --local-dir Qwen3.6-27B-GGUF |
| IQ4_XS | 14.4 GBfile size | 15.4 GB14.4 + 0.50 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.6-27B-GGUF:IQ4_XSOllama ollama run hf.co/unsloth/Qwen3.6-27B-GGUF:IQ4_XSDownload hf download unsloth/Qwen3.6-27B-GGUF --include "Qwen3.6-27B-IQ4_XS.gguf" --local-dir Qwen3.6-27B-GGUF |
| Q4_0 | 14.7 GBfile size | 15.7 GB14.7 + 0.50 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.6-27B-GGUF:Q4_0Ollama ollama run hf.co/unsloth/Qwen3.6-27B-GGUF:Q4_0Download hf download unsloth/Qwen3.6-27B-GGUF --include "Qwen3.6-27B-Q4_0.gguf" --local-dir Qwen3.6-27B-GGUF |
| Q4_K_S | 14.8 GBfile size | 15.8 GB14.8 + 0.50 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.6-27B-GGUF:Q4_K_SOllama ollama run hf.co/unsloth/Qwen3.6-27B-GGUF:Q4_K_SDownload hf download unsloth/Qwen3.6-27B-GGUF --include "Qwen3.6-27B-Q4_K_S.gguf" --local-dir Qwen3.6-27B-GGUF |
| IQ4_NL | 15.0 GBfile size | 16.0 GB15.0 + 0.50 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.6-27B-GGUF:IQ4_NLOllama ollama run hf.co/unsloth/Qwen3.6-27B-GGUF:IQ4_NLDownload hf download unsloth/Qwen3.6-27B-GGUF --include "Qwen3.6-27B-IQ4_NL.gguf" --local-dir Qwen3.6-27B-GGUF |
| Q4_K_M | 15.7 GBfile size | 16.7 GB15.7 + 0.50 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.6-27B-GGUF:Q4_K_MOllama ollama run hf.co/unsloth/Qwen3.6-27B-GGUF:Q4_K_MDownload hf download unsloth/Qwen3.6-27B-GGUF --include "Qwen3.6-27B-Q4_K_M.gguf" --local-dir Qwen3.6-27B-GGUF |
| Q4_1 | 16.1 GBfile size | 17.1 GB16.1 + 0.50 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.6-27B-GGUF:Q4_1Ollama ollama run hf.co/unsloth/Qwen3.6-27B-GGUF:Q4_1Download hf download unsloth/Qwen3.6-27B-GGUF --include "Qwen3.6-27B-Q4_1.gguf" --local-dir Qwen3.6-27B-GGUF |
| UD-Q4_K_XL | 16.4 GBfile size | 17.4 GB16.4 + 0.50 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.6-27B-GGUF:UD-Q4_K_XLOllama ollama run hf.co/unsloth/Qwen3.6-27B-GGUF:UD-Q4_K_XLDownload hf download unsloth/Qwen3.6-27B-GGUF --include "Qwen3.6-27B-UD-Q4_K_XL.gguf" --local-dir Qwen3.6-27B-GGUF |
| Q5_K_S | 17.7 GBfile size | 18.7 GB17.7 + 0.50 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.6-27B-GGUF:Q5_K_SOllama ollama run hf.co/unsloth/Qwen3.6-27B-GGUF:Q5_K_SDownload hf download unsloth/Qwen3.6-27B-GGUF --include "Qwen3.6-27B-Q5_K_S.gguf" --local-dir Qwen3.6-27B-GGUF |
| Q5_K_M | 18.2 GBfile size | 19.2 GB18.2 + 0.50 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.6-27B-GGUF:Q5_K_MOllama ollama run hf.co/unsloth/Qwen3.6-27B-GGUF:Q5_K_MDownload hf download unsloth/Qwen3.6-27B-GGUF --include "Qwen3.6-27B-Q5_K_M.gguf" --local-dir Qwen3.6-27B-GGUF |
| UD-Q5_K_XL | 18.7 GBfile size | 19.7 GB18.7 + 0.50 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.6-27B-GGUF:UD-Q5_K_XLOllama ollama run hf.co/unsloth/Qwen3.6-27B-GGUF:UD-Q5_K_XLDownload hf download unsloth/Qwen3.6-27B-GGUF --include "Qwen3.6-27B-UD-Q5_K_XL.gguf" --local-dir Qwen3.6-27B-GGUF |
| Q6_K | 21.0 GBfile size | 22.0 GB21.0 + 0.50 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.6-27B-GGUF:Q6_KOllama ollama run hf.co/unsloth/Qwen3.6-27B-GGUF:Q6_KDownload hf download unsloth/Qwen3.6-27B-GGUF --include "Qwen3.6-27B-Q6_K.gguf" --local-dir Qwen3.6-27B-GGUF |
| UD-Q6_K_XL | 23.9 GBfile size | 24.9 GB23.9 + 0.50 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.6-27B-GGUF:UD-Q6_K_XLOllama ollama run hf.co/unsloth/Qwen3.6-27B-GGUF:UD-Q6_K_XLDownload hf download unsloth/Qwen3.6-27B-GGUF --include "Qwen3.6-27B-UD-Q6_K_XL.gguf" --local-dir Qwen3.6-27B-GGUF |
| Q8_0 | 26.6 GBfile size | 27.6 GB26.6 + 0.50 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.6-27B-GGUF:Q8_0Ollama ollama run hf.co/unsloth/Qwen3.6-27B-GGUF:Q8_0Download hf download unsloth/Qwen3.6-27B-GGUF --include "Qwen3.6-27B-Q8_0.gguf" --local-dir Qwen3.6-27B-GGUF |
| UD-Q8_K_XL | 32.9 GBfile size | 33.9 GB32.9 + 0.50 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.6-27B-GGUF:UD-Q8_K_XLOllama ollama run hf.co/unsloth/Qwen3.6-27B-GGUF:UD-Q8_K_XLDownload hf download unsloth/Qwen3.6-27B-GGUF --include "Qwen3.6-27B-UD-Q8_K_XL.gguf" --local-dir Qwen3.6-27B-GGUF |
| BF16 | 50.1 GBfile size, 2 parts | 51.1 GB50.1 + 0.50 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.6-27B-GGUF:BF16Download hf download unsloth/Qwen3.6-27B-GGUF --include "BF16/Qwen3.6-27B-BF16-*-of-00002.gguf" --local-dir Qwen3.6-27B-GGUF |
MLX for Apple silicon
| Format | Size | Memory needed | Repo |
|---|---|---|---|
| MLX 4bitlmstudio-community/Qwen3.6-27B-MLX-4bit | 15.0 GBfile size | 16.0 GB15.0 + 0.50 + 0.50 | Download page |
| MLX 8bitlmstudio-community/Qwen3.6-27B-MLX-8bit | 27.5 GBfile size | 28.5 GB27.5 + 0.50 + 0.50 | Download page |
Original weights size from the safetensors header, for vLLM, SGLang, Transformers
| Format | Size | Memory needed | Repo |
|---|---|---|---|
| BF16original weights | 51.7 GBfile size | 52.7 GB51.7 + 0.50 + 0.50 |
hf download Qwen/Qwen3.6-27B --local-dir Qwen3.6-27BOther conversions (16; sizes not read)
- Qwen/Qwen3.6-27B-FP8 FP8, by the maker, 1.2M downloads
- Lorbus/Qwen3.6-27B-int4-AutoRound Other, int4, 693k downloads
- FenomAI/Qwen3.6-27B-AWQ-INT4 AWQ, int4, 379k downloads
- cyankiwi/Qwen3.6-27B-AWQ-INT4 AWQ, int4, 269k downloads
- Alennnndy/Qwen3.6-27B-AWQ-4bit AWQ, 4bit, 257k downloads
- QuantTrio/Qwen3.6-27B-AWQ AWQ, 132k downloads
- unsloth/Qwen3.6-27B-NVFP4 NVFP4, 116k downloads
- nvidia/Qwen3.6-27B-NVFP4 NVFP4, 94k downloads
- sakamakismile/Qwen3.6-27B-Text-NVFP4-MTP NVFP4, 70k downloads
- btbtyler09/Qwen3.6-27B-GPTQ-4bit GPTQ, 4bit, 45k downloads
- mmangkad/Qwen3.6-27B-NVFP4 NVFP4, 3.7k downloads
- Avesed/Qwen3.6-27B-INT8-W8A8 Other, int8, 3.3k downloads
- Intel/Qwen3.6-27B-int4-AutoRound Other, int4, 1.9k downloads
- neroued/Qwen3.6-27B-NInfer Other, 1.7k downloads
- groxaxo/Qwen3.6-27B-GPTQ-Pro-4bit GPTQ, 4bit, 603 downloads
- RedHatAI/Qwen3.6-27B-FP8 FP8, 526 downloads
The Hub tags 100 or more quantised repos as conversions of this model; we read the most downloaded and show the maker's own first.
Run it by API
5 providers serve this model through the Hugging Face router. Prices are US dollars per million tokens. Precision is what the provider reports; a lower precision than the original weights can change answers. Live first, then cheapest output.
| Provider | In | Out | Context | Precision | Tokens/s | First token | Uptime 24h | Tools | JSON |
|---|---|---|---|---|---|---|---|---|---|
| DeepInfra | $0.32 | $3.20 | 256k | fp8 | 50 | 0.5 s | 99.6% | yes | yes |
| SiliconFlow | $0.30 | $3.20 | 256k | fp8 | 29 | 1.5 s | 95.0% | yes | yes |
| Venice | $0.33 | $3.25 | 250k | fp8 | 47 | 0.7 s | 99.2% | yes | yes |
| Reka | $0.60 | $6.00 | 256k | not given | 53 | 1.1 s | 99.8% | yes | yes |
| featherless-ai | ? | ? | ? | not given | ? | ? | ? | not given | not given |
Through the router, with a Hugging Face token (the router bills your Hugging Face account). Change the name after the colon to pick another provider.
curl https://router.huggingface.co/v1/chat/completions \
-H "Authorization: Bearer $HF_TOKEN" -H "Content-Type: application/json" \
-d '{"model": "Qwen/Qwen3.6-27B:deepinfra", "messages": [{"role": "user", "content": "Hello"}]}'Not built yet: one key for every open model, with BlockRun, and our own tested-quality score per provider. Until then, precision and measured speed are the best guide to a provider here.
Fine-tunes of this model
Repos that declare this model as their base, most downloaded first. Listing is not a recommendation; unreviewed ones have not been checked.
- prism-ml/Ternary-Bonsai-27B-mlx-2bit1.0MUnreviewed
- prism-ml/Bonsai-27B-mlx-1bit996kUnreviewed
- AEON-7/Qwen3.6-27B-AEON-Ultimate-Uncensored-BF1627.4B70kUnreviewed
- Qwen/Qwen-Image-Bench27.4B18kUnreviewed
- win10/EVE-27b-XENO-HAT-DeepSeek-V4-Flash27.8B14kUnreviewed
- unsloth/Qwen3.6-27B27.8B2.9kUnreviewed
- huihui-ai/Huihui-Qwen3.6-27B-abliterated27.8B2.7kUnreviewed
- inclusionAI/ZwZ2-27B27.4B1.5kUnreviewed
- aisingapore/Qwen-SEA-LION-v4.5-27B-IT27.8B1.0kUnreviewed
- tencent/UI-Mate-27B27.4B571Unreviewed
- tencent/UI-Mate-democua-27B27.4B331Unreviewed
Planned for this page Not built yet
- Try it in private chat
- A link to Solana AI's confidential chat for the models it serves. This model is not listed as served there.
- Fine-tune it
- Fine-tuning a few base models from this page and publishing the result here.
- Creators and earnings
- Pages for people who publish fine-tunes and adapters of this model, and what they earn.
- Fund and bounties
- Bounties for new open models, paid from a public fund. How the fund will work.
- Contribute compute
- How to give compute to open model work, with Akash.
Model data from the Hugging Face Hub 19 Nov 2026, 13:43 UTC. The repo. Found a mistake on this page? Tell us.